Proc. Nat. Acad. Sci. USA Vol. 72, No. 6, pp. 2145-2149, June 1975

Genetic Variation As a Test of (allozymes/heterozygosity/environmental heterogeneity) EVIATAR NEVO*, HERBERT C. DESSAUERt, AND KUO-CHIIAN CHUANGt * Department of , University of Haifa, Haifa, Israel; and t Department of Biochemistry, Louisiana State University Medical Center, New Orleans, La. 70112 Communicated by Theodosius Dobzhansky, February 6, 1976 ABSTRACT Allozymic variation encoded by 26 loci was 27,C); (7) Nafah (n = 24,C); (8) Jericho (n = 58,1); (9) analyzed electrophoretically in 54f7 specimens representing El-Arish (n = 46,1); (10) Mashabei-Sade (n 56,1); (11) 12 populations of green toads, Bufo viridis, in Israel and the Vis Adriatic Island. in Bufo viridis is Ein-Ovdat (n = 53,I); and (12) Vis Adriatic Island (n = higher than in any vertebrate yet studied. Mean hetero- 10,1). (Populations 1-11 are from Israel.) zygosity per per individual (H) is 0.133 (range, 0.105 The Israeli populations of Bufo viridis live under extreme to 0.159). H is higher in central populations as compared ecological conditions in both space and time. Spatially, they with isolates, and varies among four major protein classes, range along ecological gradients of increasing aridity south- being highest in transferases and hydrolases and lowest in and oxidoreductases and nonenzymatic proteins. Differential wards as well as eastwards, approaching the southern frequencies among polymorphisms was tested as an eastern deserts surrounding Israel. While annual precipitation indicator of natural selection. Significant heterogeneity in the northern populations of Quneitra, Nafah, Dalton, and between loci in their apparent coefficients Saar is 850, 850, 700, and 650 mm, respectively, it decreases F. = 2p/p( -Tp) was found for all and for each of to arid desert conditions in the southern populations of El- the four major classes of proteins tested, which may be taken as evidence of selection. Both uniform and diversify- Arish, Mashabei-Sade, and Ein-Ovdat, which receive annual ing selection are suggested by the low and high F, values, precipitations of 100, 100, 75, and 65 mm, respectively. respectively. The general pattern of high heterozygosity Temporally, all Israeli populations live either under Medi- in Bufo viridis is best explained as an adaptive strategy in terranean or desert climates involving mild wet winters but heterogeneous environments. dry hot summers. Average temperatures across the range of The pattern of genetic variation within and between popula- the sampled populations are for the entire year 19.10 (range, tions may be used to test directly the alternative theories of 16-24°); for the coldest month (January) 10.50 (range, the "balanced" versus the "neo-classical" or "neutral" 6-15°), and for the hottest month (August) 25.8° (range, schools of evolutionary (1-4). While natural selec- 22-31°). Bufo tadpoles metamorphose in spring and early tion operates differentially on each , the breeding struc- summer and are thus exposed in Israel to adverse and dry ture, involving random genetic drift, inbreeding, assortative environments. Adult green toads are active during summer mating, and migration, affects all alleles similarly. Differential nights and walk long distances in search of food or mates in genic variations among polymorphisms in space and time the breeding season, thus being also exposed to environmental may thus provide evidence of selection in natural populations extremes (5). (3,4). Genetic variation may be expressed as the variance (S2) We present evidence based on differential gene frequencies of allele frequencies across populations. To overcome the of 12 populations of green toads, Bufo viridi8, in Israel and effect of the mean gene frequency (p) in computing the on the Vis Adriatic Island (H. C. Dessauer, E. Nevo, and variance, the latter may be standardized, as in the estimate K. C. Chuang, submitted for publication), suggesting that e= S2p/P (1 - P) known as the Wahlund variance (6), natural selection is the major operating evolutionary force or the "effective inbreeding coefficient" (7). (See also ref. 3 causing population differentiation. Furthermore, green toads for the statistical properties of Pe and for a sampling theory probably demonstrate an adaptive strategy for high hetero- for testing heterogeneity in ke.) Selectively neutral alleles zygosity in accord with their ecologically variable range in will all have similar estimated Pe values in spite of their space and time. variation in S2p and p because effective inbreeding will be identical for all across populations. In that neutral case, the average Pe will thus estimate the true Pe without MATERIALS AND METHODS significant heterogeneity among alleles. Lewontin and Kra- Allozymic variation encoded by 26 loci was analyzed electro- kauer (3) have showed that the theoretical variance for a phoretically in 507 adult specimens representing 12 popula- sampling distribution of Pe values from alleles not subject to tions (5 central, 2 marginal, and 5 isolates) of green toads, selection is approximately u2 = (KF2)/(n - 1) where F is Bufo viridie, 11 populations in Israel and 1 population in the the mean Fe value, n is the number of populations under Vis Adriatic Island. The 12 populations, sample sizes, and study, and K = 2, assuming a binomial distribution of allele their geographical location, central (C), marginal (M), and frequencies. If, however, selection is operating on some or all isolate (I), are as follows: (1) Saar (n = 73,C); (2) Tel-Aviv of the loci, their Pe values will be significantly heterogeneous (n = 28,C); (3) Ein Hashlosha (n = 53,M); (4) Dalton and will not be estimates of the same Pe. The ratio of the (n = 47,C); (5) Jerusalem (n = 42,M); (6) Quneitra (n = observed variance of Pe's (S2F) to the theoretical variance 2145 Downloaded by guest on September 27, 2021 2146 Genetics: Nevo et al. Proc. Nat. Acad. Sci. USA 72 (1975)

TABLE 1. Gene frequencies and effective inbreeding coefficients of alleles at 20 polymorphic loci in populations of Bufo viridis*

Ein Saar Tel- Hash- Jeru- El- Mashabei Ein- With- n = Aviv losha Dalton salem Quneitra Nafah Jericho Arish Sade Ovdat Vis out With Locust Allele (73)1t (28) (53) (47) (42) (27) (24) (58) (46) (56) (53) (10) Vis Vis

I. O 1i9oreductases Ldh ba 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.991 1.000 1.000 1.000 1.000 0.009 0.009 b 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.009 0.009 Ldh-2 a 0.000 0.000 0.000 0.011 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.011 0.011 b 0.281 0.429 0.226 0.298 0.048 0.278 0.208 0.231 0.272 0.241 0.047 0.000 0.064 0.091 c 0.719 0.571 0.774 0.691 0.952 0.722 0.792 0.769 0.728 0.759 0.953 1.000 0.067 0.072 Mdh-i a 1.000 1.000 1.000 1.000 0.988 1.000 0.958 1.000 1.000 1.000 1.000 1.000 0.033 0.034 b 0.000 0.000 0.000 0.000 0.012 0.000 0.042 0.000 0.000 0.000 0.000 0.000 0.034 0.034 Icd-i a 0.058 0.019 0.028 0.021 0.115 0.065 0.043 0.070 0.000 0.018 0.000 0.000 0.032 0.035 b 0.507 0.685 0.519 0.426 0.500 0.435 0.391 0.480 0.598 0.670 0.702 0.650 0.048 0.048 c 0.435 0.296 0.453 0.553 0.385 0.500 0.565 0.450 0.402 0.313 0.298 0.350 0.037 0.036 Icdc2 a 0.021 0.036 0.009 0.000 0.000 0.038 0.000 0.000 0.000 0.009 0.000 0.000 0.021 0.022 b 0.979 0.928 0.991 1.000 1.000 0.962 1.000 1.000 1.000 0.991 1.000 1.000 0.039 0.040 c 0.000 0.036 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.036 0.036 6Pgd a 0.979 0.964 0.981 0.947 1.000 0.944 0.979 0.991 0.989 1.000 1.000 1.000 0.020 0.022 b 0.021 0.036 0.019 0.053 0.000 0.056 0.021 0.009 0.011 0.000 0.000 0.000 0.020 0.022 Ipo a 0.000 0.000 0.000 0.000 0.024 0.000 0.000 0.034 0.000 0.000 0.000 1.000 0.027 1.027 b 1.000 1.000 1.000 1.000 0.976 1.000 1.000 0.966 1.000 1.000 1.000 0.000 0.027 1.027

II. Transferases Got-i a 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.008 0.008 b 0.959 0.982 0.982 0.947 1.000 0.963 1.000 0.991 1.000 1.000 0.981 0.000 0.020 0.996 c 0.041 0.018 0.009 0.053 0.000 0.037 0.000 0.000 0.000 0.000 0.019 0.000 0.024 0.025 d 0.000 o.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.000 0.000 1.091 Got-2 a 0.007 0.000 0.009 0.021 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.013 0.014 b 0.972 1.000 0.991 0.979 1.000 1.000 1.000 1.000 1.000 1.000 0.981 1.000 0.019 0.017 c 0.021 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.019 0.000 0.018 0.018 Gpt-2 a 0.492 0.620 0.417 0.345 0.431 0.452 0.548 0.417 0.500 0.430 0.549 1.000 0.024 0.115 b 0.508 0.380 0.583 0.655 0.569 0.548 0.452 0.583 0.500 0.570 0.451 0.000 0.024 0.115 Pgm2 a 0.007 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.007 b 0.387 0.393 0.398 0.167 0.313 0.405 0.250 0.321 0.239 0.295 0.368 0.000 0.028 0.068 c 0.585 0.607 0.602 0.782 0.688 0.595 0.729 0.679 0.761 0.682 0.585 1.000 0.024 0.067 d 0.021 0.000 0.000 0.051 0.000 0.000 0.021 0.000 0.000 0.023 0.047 0.000 0.026 0.027

III. Hydrolases Est-5 a 0.040 0.000 0.074 0.070 0.013 0.096 0.074 0.063 0.013 0.028 0.019 0.000 0.024 0.028 b 0.500 1.000 0.596 0.302 0.825 0.519 0.593 0.313 0.923 0.594 0.481 1.000 0.294 0.260 c 0.397 0.000 0.330 0.267 0.038 0.385 0.222 0.328 0.064 0.104 0.327 0.000 0.124 0.146 d 0.048 0.000 0.000 0.337 0.113 0.000 0.111 0.266 0.000 0.274 0.173 0.000 0.149 0.125 e 0.016 0.000 0.000 0.023 0.013 0.000 0.000 0.031 0.000 0.000 0.000 0.000 0.017 0.018 Est-6 a 0.462 0.407 0.441 0.271 0.338 0.240 0.313 0.345 0.322 0.357 0.491 0.389 0.027 0.025 b 0.456 0.519 0.441 0.630 0.618 0.680 0.563 0.582 0.678 0.571 0.472 0.444 0.029 0.031 c 0.083 0.074 0.118 0.098 0.044 0.080 0.125 0.073 0.000 0.071 0.038 0.167 0.019 0.026 Ap-i a 0.027 0.000 0.000 0.000 0.012 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.017 0.017 b 0.952 1.000 1.000 1.000 0.976 1.000 0.979 0.982 1.000 0.991 1.000 1.000 0.022 0.023 c 0.021 0.000 0.000 0.000 0.012 0.000 0.021 0.009 0.000 0.009 0.000 0.000 0.011 0.012 Ap-2 a 0.993 0.926 0.736 1.000 1.000 1.000 0.958 1.000 1.000 0.991 1.000 1.000 0.180 0.182 b 0.007 0.074 0.264 0,000 0.000 0.000 0.042 0.000 0.000 0.009 0.000 0.000 0.181 0.182 Pep-2 a 0.384 0.188 0.208 0.255 0.513 0.185 0.271 0.267 0.022 0.145 0.094 0.050 0.102 0.113 b 0.616 0.813 0.792 0.745 0.487 0.815 0.729 0.733 0.978 0.855 0.906 0.950 0.102 0.113 Pro a 0.014 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.018 0.009 0.000 0.012 0.012 b 0.986 1.000 0.991 1.000 1.000 1.000 1.000 1.000 1.000 0.982 0.991 1.000 0.010 0.010 c 0.000 0.000 0.009 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.009 0.009

(Continued on facing page.) Downloaded by guest on September 27, 2021 Proc. Nat. Acad. Sci. USA 72 (1975) Genetic Variation as a Test of Natural Selection 2147 TABLE 1. (continued)

Ein Saar Tel- Hash- Jeru- El- Mashabei Ein- With- n = Aviv losha Dalton salem Quneitra Nafah Jericho Arish Sade Ovdat Vis out With Locust Allele (73)1 (28) (53) (47) (42) (27) (24) (58) (46) (56) (53) (10) Vis Vis IV. Nonenzymatic proteins Alb a 0.007 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.007 b 0.986 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.018 0.014 c 0.007 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.007 0.007 TfT a 0.508 0.587 0.510 0.614 0.375 0.611 0.636 0.526 0.350 0.404 0.179 0.080 0.080 b 0.492 0.413 0.490 0.386 0.625 0.389 0.364 0.474 0.650 0.596 0.821 0.080 0.080 Pa-2 a 0.014 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.014 0.014 b 0.986 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 0.014 0.014

* Abbreviations: ofGpd: glycerol-3-phosphate dehydrogenase (EC 1.1.1.8); Gdh: glutamate dehydrogenase; Ldh-2: lactate dehydro- genase-2; Mdh-i: malate dehydrogenase-1; Icd-i and -2: isocitrate dehydrogenase-1 (EC 1.1.1.41) and -2 (EC 1.1.1.42); 6Pgd: 6-phospho- gluconate dehydrogenase-1 (EC 1.1.1.43); Cnp: cyanoperoxidase (EC 3.5.5.3); Ipo: indophenol oxidase-1 (EC 1.9.3.1); Got-i and -2: glutamic oxaloacetic transaminase-1 and -2 (EC 2.6.1.1); Gpt-2: glutamate pyruvic transaminase-2 (EC 2.6.1.2); Pgm-2: phosphogluco- mutase-2(EC 2.7.5.1); Est-5 and -6: esterase-5 and -6; Ap-i and -2: acid phosphatase-1 and -2 (EC 3.1.3.2); Pep-2: peptidase-2; Pro: prolidase-1 (EC 3.4.13.9); Alb: albumin; Tf: transferrin; Pa-i and -2: proalbumin-1 and -2. t aGpd, Gdh, Cnp, Gpt-i, Pep-i, and Pa-i were invariant in all samples. $ Number of specimens. § Uncorrected for sampling error. ¶ No data on Vis Island sample.

(a2) is distributed as x2, the degrees of freedom (d.f.) referring populations examined. The three parameters in Vis are 1.12, to one less than the number of Pe values. Thus, x2 d.f. = 0.12, and 0.029, respectively. (d.f.) (S2F/02) tests the observed variance of the distribution The pattern of allelic variation in Bufo Piridis has the follow- of Pe values against its expected value in the absence of ing features: (a) Of the 26 loci examined, seven (aGpd, Ldh-i, selection, and it will be large if the distribution of Pe's is Gdh, Cnp, Gpt-i, Pep -1, and Pa-i) are essentially mono- heterogeneous. The test of homogeneity of Pe estimates from morphic, and homogeneous from the Vis Adriatic Island to different loci in steady state is thus a test for selection. To the Negev and Sinai deserts in southern Israel. (b) The 20 correct for sampling error resulting from the variance in allele polymorphic loci involve 12 weakly polymorphic (Ldh-i, frequencies within each sampled population, particularly in Mdh-1, Icd-2, 6Pgd, Ipo, Got-i, Got-2, Ap-1, Ap-2, Pro, small populations, we subtract 1/2 A from the observed Fe Alb, and Pa-2) and 8 strongly and regionally polymorphic loci values, calculated from the above-mentioned equation, where (Ldh-2, Icd-i, Gpt-2, Pgm-2, Est-5, Est-6, Pep-2, and Tf). n is the average population size on which the (c) In the weakly polymorphic loci, and in some strongly estimates are based (see also refs. 2 and 6). polymorphic ones (Ldh-2, Icd-2, and Got-i, except Vis, Pgm-2 and Pep-2), one allele is mostly predominant in fre- RESULTS quency across all populations, including the desert geographi- The allelic frequencies of the 20 polymorphic loci in the 12 cal isolates. (d) The similarity of all allozyme distributions populations of Bufo viridis with their respective Pe values across Israel is reflected by high genic similarities (8), varying are given in Table 1. Alleles are designated alphabetically in from 0.93 to 0.97 and averaging 0.95. (e) Two polymorphic order of decreasing mobility of their allozymes. The loci in loci (Icd-i and Tf) exhibit clinal patterns significantly cor- Table 1 are arranged in four major classes: oxidoreductases, related (0.02 > P > 0.01) with increasing aridity. (f) In three transferases, hydrolases, and nonenzymatic proteins. loci (Gpt-2, Est-5, and Est-6) two alleles predominate across Genetic variation in Bufo viridis is higher than in any the range. (g) Alternative fixation is nonexistent in the vertebrate yet studied. All three genetic parameters, mean Israeli mainland populations, and in the sole case present, in numbers of alleles per locus (A), mean proportion of loci Vis Adriatic Island, it occurs in only 2 loci out of 26 (Ipo polymorphic per population (P), and mean proportion of and Got). (h) Finally, of major classes of enzymes, genetic loci heterozygous per individual (H) are very high (A = 1.65, variation was least in oxidoreductases (A = 1.9; P 0.40; range: 1.38-2.04; P = 0.423, range: 0.385-0.615; and H = H = 0.091); next come nonenzymatic proteins (A = 2.0; 0.133, range: 0.105-0.159). Genetic variation (in both P and P = 0.25; H = 0.103) followed by hydrolases (A = 2.7; H) is significantly lower (P < 0.05) in Israeli isolates as com- P = 0.71; H = 0.155) and transferases (A = 2.6; P = 0.60; pared with central and marginal populations. The three H= 0.218). genetic parameters A, P, and H are 1.69, 0.47, and 0.14, The P, values (Table 1) are distinctly heterogeneous from respectively, in central populations; they are 1.67, 0.48, and a low of 0.011 to a high of 0.294, excluding Vis, and 0.011- 0.14, respectively, in marginal populations; and 1.58, 0.37, 1.09 including Vis. This apparent heterogeneity of Ae was and 0.12, respectively, in Israeli isolated populations. Vis tested by comparing the observed (corrected for sampling Adriatic Island is by far the least polymorphic of all the error) and theoretical variances (S2F/o2) for all the alleles of Downloaded by guest on September 27, 2021 2148 Genetics: Nevo et al. Proc. Nat. Acad. Sci. USA 72 (1975)

TABLE 2. Summary of four tests for the heterogeneity of P. by comparing observed and theoretical variances of P. for all alleles and four groups of proteins in 11 Israeli populations of Bufo viridis Effective inbreeding coefficient Critical F Observed* value with Alleles Theoretical d.f. t Group (n) Mean(F.) Variance(S2F) (o2) S2F/o2 (F 0.001 d.f.) P All alleles 55 0.033 0.00286 0.00021 13.49 (35, co) = 1.99 (<0.001 Oxidoreductases 17 0.022 0.00028 0.00009 3.01 (10, x) = 2.96 <0.001 Transferases 13 0.010 0.00004 0.00002 2.49 (9, oo) = 3.10 0.01-0.005 Hydrolases 18 0.064 0.00681 0.00082 8.34 (12, c) = 2.74 <<0.001 Nonenzymatic proteins 7 0.022 0.00107 0.00010 10.77 (3, cc) = 5.42 (<0.001

* Corrected for sampling error by subtracting 1/2FI from the observed P.; mean n = 46. t Subtracting a degree of freedom (d.f.) for each multiple allele locus and for each di-allelic locus (3). all 20 polymorphic loci, then for each of the above-mentioned the mesic-xeric range of green toads in Israel negate the drift protein groups separately. The results are given in Table 2. hypothesis and suggest natural selection. Selection of increas- The number of populations in the analysis is n = 12, but only ing body size of green toads southwards, correlated with the the 11 Israeli mainland populations were analyzed for the Pe same climatic gradient of increasing aridity, has been de- test. From this, the theoretical variance = (KF2)/(n - 1) scribed by Nevo (9). was calculated by a2 = 2.0 F2/10, and P. or mean Pe is given Natural selection is further suggested as the major deter- for each group separately in Table 2. The ratios between the minant of genetic variation in green toads, by testing the observed and theoretical variances of inbreeding coefficients distribution of gene frequency directly (3). Significant hetero- is also given in Table 2, together with the critical values of the geneity between loci in their apparent inbreeding coefficients, F distribution, which is distributed as X2/d.f. (3). Fe ranging from 0.011 to 0.294, was found for 55 alleles in mainland Israeli Bufo viridis, suggesting that natural selection DISCUSSION is causing the observed variation of allele frequencies. More- The pattern of genetic variation in populations of Bufo viridis over, the difference in the Fe may suggest the type of selection is inexplicable on a neutral hypothesis of protein polymor- operating (3). Uniform selection across the range will result phisms, but supports natural selection as a major evolutionary in a low variance, hence in low Fe values. This is true not only factor in population differentiation. The 11 Israeli populations for all monomorphic loci, but also for those polymorphic loci are distinctly similar genetically (3 = 0.95, range 0.93-0.97), whose alleles are either prevalent or close to fixation across yet migration can be completely ruled out to explain similari- populations (Ldh-1, Mdh-1, Icd-1, 6Pgd, Got-2, Ap-2, Pro, ties between central populations and geographical isolates in Alb, and Pa-2). On the other hand disruptive selection will the desert. While gene flow may be continuous in central and result in high variance across populations, suggesting differen- marginal populations of green toads where bodies of water tial selection in different environments. This appears to be stop in isolated popu- the case in strongly polymorphic loci displaying geographical are interconnected, it must completely P lations of the Negev and Sinai deserts. Populations such as dines in allele frequency (Icd-I and Tf ) as well as in those El-Arish, Mashabei-Sade, Ein-Ovdat, and partly Jericho are displaying alternative fixation of alleles between the Vis isolated geographically in desert oases with vast intervening Island and Israeli populations (Ipo and Got-i). The merit of dry desert areas inaccessible for toads. The high genic simi- analyzing genic patterns among polymorphisms and among larity in Israeli green toads despite ecogeographical desert populations, as compared with direct studies of viability and isolation negates a neutral hypothesis, and supports natural fertility, relates to the fact that the genetic pattern is the selection as a major evolutionary factor in population dif- accumulative results of many generations of breeding struc- ferentiation. ture and selection. Since breeding structure should affect all is neither supported alleles similarly, heterogeneity in Fe's suggests natural selec- Likewise, a random drift hypothesis studied on evidence of population structure nor on that of the genic tion, operating directly on some of the loci being drift would predict low hetero- and/or on closely linked loci. patterns. A random hypothesis in viridis zygosity as well as alternative fixations in different local The general pattern of high genic variation Bufo initial in the is best explained as an adaptive strategy for high heterozygos- populations according to the gene frequencies toads original populations. In Israeli Bufo viridis heterozygosity ity in the heterogeneous environments in which green varies geographically from 0.159, in Saar central population, live. Heterozygosity appears to be higher in organisms living Sinai of in spatially and temporally variable environments as com- to 0.105 in the small isolated Northern population In El-Arish. Remarkably, heterozygosity in the isolated desert pared to those living in relatively constant ones (10-14). than H Israel, underground mole rats (13), underground toads population of Jericho is as high as 0.140, even higher mole in several central populations such as Tel-Aviv, Dalton, and Pelobates (E. Nevo, in preparation), and underground 11 Israeli no one alterna- crickets Gryllotalpa (E. Nevo, in preparation), all living along Quneitra. Within the populations herein for tive fixation was found in 20 polymorphic systems comprising the same ecological transect described Bufo viridis, in small isolated display low heterozygosity. In contrast, the landsnail Theba 55 alleles. Both the high heterozygosity variation similar desert populations and the uniformity of many alleles across pisana is highly variable, displaying genetic Downloaded by guest on September 27, 2021 Proc.PNat...Acad.d9GeneticSci. USA 72 (1975) Variation as a Test of Natural Selection 2149

to that of Bufo viridis (15). It therefore seems plausible that Science Foundation (BSF Grant 330), to E.N. and R. C. Lewontin least and National Science Foundation of the United States (Grant the degree of genetic variation of wild populations is at BMS 73-01252 A01) to H.C.D. partly regulated by natural selection, owing to environmental variability. 1. Cavalli-Sforza, L. L. (1966) Proc. Roy. Soc. London Ser. B. 164, 362-379. Note Added in Proof: While this paper has been in press, 2. Bodmer, W., Cann, H. & Piazza, A. (1972) in Histocom- Ewens and Feldman (16) have questioned the validity of the patibilty Testing 1972, eds. Dausset, J. & Colombani, J. (Munsgaard, Copenhagen), pp. 753-767. F-test proposed by Lewontin and Krakauer (3), as well as 3. Lewontin, R. C. & Krakauer, J. (1973) Genetics 74, 175- other current tests of neutrality, on theoretical grounds. 195. They argue that tests of neutrality are not well-founded on 4. Nevo, E. (1973) Nature 244, 573-575. theory and involve internal contradictions by using in- 5. Nevo, E. (1974) Encyclopedia Hebraica (in Hebrew), in press. 6. Cavalli-Sforza, L. L. & Bodmer, W. (1971) The Genetics of appropriate mathematical models for the electrophoretic data. Human Populations (W. H. Freeman and Co., San Fran- They suggest that, while future models may improve, it is cisco). still possible that they will not have sufficient statistical 7. Lewontin, R. C. (1974) The Genetic Basis of Evolutionary power to resolve the neutrality question without larger Change (Columbia Univ. Press), Irvington-on-Hudson, amounts of data. They also point out that alternative avenues N.Y.). 8. Rogers, J. S. (1972) Univ. Tex. Pubi. 7213, 145-153. for testing neutrality are possible, involving environmental 9. Nevo, E. (1972) Ir. J. Med. Sci. 8, 1010. correlates and tests related to gene frequencies of functionally 10. Gorman, G., Soul6, M., Yang, S. Y. & Nevo, E. (1975) different enzymes. These suggestions have been independently Evolution, in press. applied to the Bufo viridi8 gene frequency data in the present 11. Levinton, J. (1973) Science 180, 75-76. C. paper, as as C. Dessauer, E. and 12. Nevo, E., Kim, Y. J., Shaw, C. R. & Thaeler, S. (1974) well in that of H. Nevo, Evolution 28, 1-23. K. C. Chuang, submitted for publication. The statistically 13. Nevo, E. & Shaw, C. R. (1972) Biochem. Genet. 7, 235-241. significant heterozygosity estimates among the four protein 14. Nevo, E. (1975) in Proc. Intern. Confer. on Population groups discussed in this paper, oxidoreductases, transferases, Genetics & Ecology 1976, eds. Karlin, S. and Nevo, E. hydrolases, and nonenzymatic proteins, are incompatible (Academic Press, New York), in press. 15. Nevo, E. & Bar. Z. (1975) in Proc. Intern. Confer. on Popu- with a neutral hypothesis and suggest that differential selec- lation Genetics & Ecology 1975, eds. Karlin, S. and Nevo, E. tion is operating at the enzyme level. (Academic Press, New York), in press. 16. Ewens, W. J. & Feldman, M. (1975) in Proc. Intern. Con- We thank Mr. M. Avrahami and Mr. G. Heth for field assis- fer. on Population Genetice & Ecology 1976, eds. Karlin, S. tance. This work was supported in part by U.S.-Israel Binational and Nevo, E. (Academic Press, New York), in press. Downloaded by guest on September 27, 2021